Young Star Clusters: Metallicities, Ages, and Masses

arXiv:astro-ph/0003417v1 28 Mar 2000 .LncnadCM ol,eds. Boily, C.M. Lan¸con and A. yMue 19) a ie nsaefo hi FsneMLvar M/L since LF their from shape in a differ YSCs, may of clo ( MF (1995), molecular The function Meurer or mass processes. by clouds formation The cluster molecular and mu of radii. of star are radii about GC MFs Effective YSCs to the Way. with similar these Milky comparison pc for the few in derived im a clusters Masses are far-reaching open of with GCs. those young question than are burning YSCs a these and remnants merger S ofrneSre,Vol. Series, Conference ASP Clusters Stellar Massive o ihms lses ia hcigadeaoainpref evaporation and shocking clusters. tidal mass fric clusters, dynamical by mass destruction high ti While for and masses. cluster processes on destruction c pend that of shows modelling potential Dynamical Galactic processes. be might destruction o but cluster not fading, by is evolutionary LF stellar observed by survivors shifted their mation hardiest Hence, un the 1991). be are (Harris to GCs population” “present-day believed hand, are to systems other distances GC the are of of They LFs determinations obj formation. Universe. for galaxy oldest the of of the times age (among) the masses? the to be YSC back to and dating believed verse, GC conventionally know are to want GCs we do Why 1. rgtYC r bevdi ag ubr nitrciggal interacting in numbers large in observed are YSCs Bright Abstract. Göttingen, Germany 11 Universitätssternwarte Alvensleben Göttingen, Geismarlandstr. v. – Fritze Uta Masses and Ages, Metallicities, Clusters: Star Young cigglxe r eiwdwt atclrepai nth on emphasis ( particular Functions with reviewed are galaxies acting ubetm oosrain fodGoua lse ( Cluster Globular old compare of be observations will to that – br systems time their future YSC Hubble of the the a predict basis of models the evolution ages on YSC luminosity YSCs individual individual With for abu colours. ages synth YSC derive Evolutionary to to models. compared low galaxy be spiral will from They predictions available. are measurements in ao ore fucranyaetemtliiy dus metallicity, mas the underlying are the o uncertainties. uncertainty e of colour shape of be observational shape the sources and can the cause Major masses can from YSC systems differ tion. age, young substantially in of to effects function LF spread a Age as ratios mated. M/L model Using LF bevtoso on trCutr( Cluster Star Young of Observations n oordsrbtos e pcrsoi abundance spectroscopic few A distributions. colour and ) 3 × 10 8 2000 , > 0Mcado h ubecntn.On constant. Hubble the of and Mpc 20 1 MF YSC rnilydsrylow destroy erentially eclssrnl de- strongly mescales diinlymodified additionally utrssesi the in systems luster fYCssesin systems YSC of ) ini oeefficient more is tion l hi Fa for- at LF their nly 37083 – D , falre original larger a of ytm ninter- in systems ) rtpitdout pointed first s sdt constrain to used dcrstlsus tells cores ud csi h Uni- the in ects Sstypically YSCs lctosi if is plications i Luminosity eir ssmdl al- models esis vra enough iversal GC reddening, t oorand colour e rapidly ies systems. ) xe and axies hhigher ch a band oad after – d ndance func- s the f sti- 2 U. Fritze – v. Alvensleben at young ages and the age spread within a YSC system is not much smaller than its age. The LF of open clusters in the Milky Way, the MFs of molecular clouds and molecular cloud cores, the observed LFs of YSCs, e.g. in NGC 4038/39, NGC 7252, NGC 3256, and of giant HII regions all are power laws with slopes in the range α ∼ −1.5 . . . − 1.8 (cf. Solomon et al. 1987, Lada et al. 1991, Kennicutt 1989, and reviews by Harris & Pudritz 1994, Elmegreen & Efremov 1997). Yet the LF of old GCs is Gaussian with typical parameters hMVi∼−7.3 mag, σ(MV) ∼ 1.3 mag, their MF is log-normal with typically hLog(M/M⊙)i∼ 5.5, σ ∼ 0.5 (e.g. Ashman et al. 1995). Hence the question as to the MF of YSCs has profound implications. If the MF of YSCs were a power law like their LF and if YSC systems are to evolve into something similar to old GC systems, dynamical destruction processes would have to transform the power law MF into a log-normal MF over a Hubble time. If, on the other hand, the MF of YSCs were log-normal (and their LF distorted to a power law by age spread effects), then the star/cluster formation process would have to transform the power law MF of the molecular clouds into the log-normal MF of YSCs. Or, else, might already the MF of molecular clouds/cloud cores in violently star forming mergers (where to my knowledge it has not yet been observed) be different from what it is in quietly star forming ‘normal galaxies’ (where it is observed)?

2. Evolutionary synthesis of SSPs

Star clusters are simple stellar populations (SSPs), formed in one short burst of star formation with one metallicity Z. With evolutionary synthesis models, the time evolution of SSPs of various metallicities Zi is studied in terms of luminosities Lλ, colours, spectrum, absorption features, stellar mass loss, and hence M/L by many groups (e.g. Bruzual & Charlot 1993, Worthey 1994, Bressan et al. 1994, F.-v.A. & Burkert 1995, Leitherer et al. 1999, Kurth et al. 1999). Ba- sic parameters of this approach are the IMF and the set of stellar evolutionary tracks used (e.g. from the Padova or Geneva groups). All models agree that the changes in luminosities and colours are rapid in early evolutionary stages and become slower with increasing age. The colour evolution depends signiﬁ- cantly on metallicity, already in very early stages, and the fading also depends on metallicity, in particular during the ﬁrst Gyr (cf. F.-v.A. & Burkert 1995, Kurth et al. 1999).

3. YSC observations vs. SSP models

3.1. Procedure If the metallicity of YSCs is known, individual ages can be obtained on the basis of their observed UBVI colours. Unfortunately, metallicity information from spectroscopy is only available yet for a handful of YSCs in NGC 7252 (Schweizer & Seitzer 1993, 1998) and NGC 1275 (Brodie et al. 1998). In all other cases we have to go back to some educated guess. In gas rich galaxy mergers, YSCs form out of the gas from the progenitor (spiral) galaxies, the abundance of which gives a lower limit to YSC abundances. E.g., for the YSCs Star Clusters: Metallicities, Ages, and Masses 3 in NGC 7252, which is a merger of two luminous Sc type spirals, we had predicted Sc 1 ZYSC ∼> ZISM ∼ /2Z⊙ from the ISM abundance evolution in our 1-zone spiral models (F.-v.A. & Gerhard 1994). Spectroscopy of the brightest YSCs yielded ZYSC ∼ Z⊙ with some tentative indication of self-enrichment during the burst from the comparison of Mgb and Fe lines (cf. F.-v.A. & Burkert 1995). Distances are known for all YSC systems. Hence absolute luminosities LV can be combined with M/LV− values from SSP models of appropriate age and metallicity to derive the masses of YSCs.

3.2. Sources of uncertainty Sources of uncertainty for these mass estimates come both from observations and models. Observational errors on luminosities and colours might be inhomogeneous and probably not independent of each other, dust extinction is not known for individual YSCs, the dust distribution in some systems is inhomogeneous, the metallicity is, at most, known for a small subsample of YSCs and might show an intrinsic scatter, and, finally, the completeness limit need not be homogeneous over the region of YSC observations. Intrinsic model uncertainties are estimated to be ∼< 0.1 mag in (optical) luminosities and colours. Differences between models from various authors are mainly due to differences in the stellar input physics. While serious colour discrepancies at very young ages ∼ 10 Myr are seen, e.g. comparing models from Bruzual & Charlot with those of Leitherer et al. , probably due to the inclusion/non-inclusion of emission lines, fairly good agreement is reached among models from various groups for the same metallicity and IMF at all ages ∼> 60 − 100 Myr. E.g., ∆(V − I) |12 Gyr∼< 0.1 mag and ∆(M/LV) |12 Gyr∼< 10%. M/Lλ, however, depends on wavelength λ, metallicity Z, and IMF, i.e. on its slope and lower mass limit. In Tab.1, I briefly sketch out these dependencies as obtained from our models using Padova stellar evolutionary tracks for stars in the mass range 0.1 − 60 M⊙ (cf. Kurth et al. 1999) and two different IMFs (Salpeter vs. Scalo 1986). In general and as reported by others before, model M/LV are about twice as large as are M/LV− values derived from observations of old GCs, even when a Scalo IMF is assumed. For a Salpeter IMF the discrepancy is higher by another factor of two. Observational M/L− values are obtained by measuring the central velocity dispersion σ0, central surface brightness I0, and half light radius r1/2, and assuming isotropic orbits, no radial gradients in M/L, and no DM halos around GCs. Then 2 M/L ∼ σ0 I0 r1/2 and typical values quoted for Galactic GCs are M/LV ∼ 2. There are two effects that might invalidate the assumptions going into this derivation. First of all, mass segregation in GCs (cf. Meylan this conf.) will result in an M/L increasing with radius, as e.g. observed by Côté et al. 1995 for NGC 3201. Second, low mass stars with their high M/L− values are preferentially lost by evaporation (e.g. Gerhard this conf.). While both processes are undoubtedly at work in GCs, attempts to examine quantitatively the validity of the assumptions involved seem to still give ambiguous results. Leonard et al. 1992 use proper motion data and radial velocity measurements for stars in the GC M13 and find that ∼ 50% of 4 U. Fritze – v. Alvensleben the mass of M13 is in low mass stars and brown dwarfs. Including this unseen mass leads to an increased M/L ∼ 4, a value well compatible with the models. On the other hand, observations of tidal tails on some GCs in the Milky Way potential are interpreted to indicate that there is not much DM around those GCs, constraining their mass-to-light ratios to M/L ∼< 2.5 (Moore 1996).

Table 1. M/L-values from SSP models at various wavelengths for young, intermediate age, and old stellar populations compared for two metallicities and two diﬀerent IMFs.

−3 Z = 10 2 · Z⊙

8 10 yr M/LU,B ∼ 0.1 0.1 indep. of M/LV,R ∼ 0.2 0.2 Z & IMF M/LK ∼ 0.3 0.5 indep. of IMF

12 Gyr M/LB,V |Salp ∼ 8 12 ∼ 2 × M/LB,V |Scalo M/LK |Salp ∼ 5 1.5 – M/LK |Scalo ∼ 3.5 2.7 –

4. YSCs in the Antennae: a 1st example

The Antennae galaxies (= NGC 4038/39) is an interacting pair of two gas rich spirals, probably Sc, of comparable mass. In the ongoing starburst triggered by the interaction a population of bright YSCs is formed, numerous enough to allow for the first time to reasonably define a LF. The LF from WFPC1 observations is a power law with slope α ∼ −1.8 (Whitmore & Schweizer 1995 (WS95)). WFPC2 reobservations are going to be presented by Miller (this conf.). Assuming a homogeneous metallicity Z ∼ 1/2 · Z⊙ lack of individual cluster spectroscopy and in analogy to the YSC system in NGC 7252, I analysed the WFPC1 data of WS95 with evolutionary synthesis models for SSPs. In a first step, the average dereddened (V − I) colour of the 550 YSCs is used to derive a mean age of the YSC population of (2±2)·108 yr, consistent with Barnes’ (1988) dynamical model for the interaction between NGC 4038 and 4039 and consistent with Kurth’s (1996) global starburst age. SSP models describe both the fading and the reddening of the YSC population as it ages. Assuming a mean age for the YSC population, both the LF and the colour distribution of the YSCs are simply shifted towards fainter magnitudes and redder colours, respectively, without changing shape. Star Clusters: Metallicities, Ages, and Masses 5

4.1. Evolution of the YSC LF In a second step, individual ages are derived for all the YSCs on the basis of their individual (V − I) and (U − V) colours. Interestingly, the resulting age distribution not only shows a peak at very young ages 0 − 4 · 108 yr for the YSCs, but also a contribution from ∼ 12 Gyr old GCs from the parent galaxies (Fig. 1a). A small number of apparent interlopers are probably due to inhomogenities in the dust distribution. It is improbable that many of the old GCs we identify are highly reddened YSCs, since they would have to be exeedingly bright intrinsically. The fact that age estimates from (U − V) colours, as far as available, agree with age estimates from (V − I) supports our metallicity assumption and makes us hope that the average EB−V is correct for most of the clusters. It is clear, however, that the individual YSC extinctions are a major source of uncertainty in this analysis. With individual YSC ages and observed luminosities, SSP models can be used to predict the time evolution of cluster luminosities. Neglecting any kind of dynamical cluster destruction effects, and only using the young star clusters identified, we obtain the surprising result that by an age of 12 Gyr, when age differences among individual YSCs will be negligi- ble, their LF will have evolved from the presently observed power law into a fairly normal Gaussian GC LF with parameters MV = −6.9 mag and σ(MV) = 1.3 mag (Fig. 1b). The turn-over then is ∼> 1 mag brighter than the completeness limit which evolves to MV = −5.7 mag, and fainter by ∼ 0.4 mag than for typical GC systems. This is a consequence of the enhanced metallicity of the YSC population with respect to that of old GC systems, and in agreement both with observations of old GC systems with a range of metallicities (Ashman et al. 1995) and with our SSP model predictions. With this surprising result we confirm and quantify Meurer’s conjecture that over a Hubble time, age spread effects can transform an observed power law LF of YSCs into the Gaussian LF of old GCs (cf. F.-v.A. 1998 for details). The bright end of the LF defined by the old GCs from the parent spirals is well described by a Gaussian LF with parameters hMVi = −7.3 mag and σ(MV) = 1.2 mag, normalised to the total number of GCs in the Milky Way and Andromeda galaxies together (Fig. 2a). Hence, if the two interacting spirals NGC 4038 and 4039 had a similar number of GCs as those galaxies, and if the bulk of the YSC population really are young GCs, Zepf & Ashman’s (1993) requirement, that the number of secondary GCs formed in mergers should be comparable to the number of primary GCs in the two progenitor spirals, would be fulfilled. This requirement is necessary if the higher specific GC frequency in ellipticals – as compared to spirals – is to be compatible with a spiral-spiral merger origin for those ellipticals.

4.2. MF of YSCs To investigate whether the LFs and MFs of old GC systems are determined by the cluster formation process or whether they are the result of secular dynamical destruction processes, we derive the MF of the very young star cluster system in the Antennae. We restrict ourselves to YSCs brighter than the completeness limit, use individual YSC ages and luminosities, and combine them with model M/L for the respective YSC ages to determine their individual masses. Our tentative conclusion is that for all the 393 YSCs brighter than the com- 6 U. Fritze – v. Alvensleben

Star Clusters in the Antennae Age Distribution 200.0

150.0

100.0 Number of Clusters

50.0

0.0 -1.0 1.0 3.0 5.0 7.0 9.0 11.0 13.0 Age [Gyr]

Figure 1. (a) Age distribution of all clusters, (b) LF evolved to 12 Gyr by diﬀerential fading for young clusters in the Antennae. The vertical arrow indicates the completeness limit.

80.0

60.0

40.0 Number of Clusters

20.0

0.0 4.0 5.0 6.0 7.0 8.0 Log M_YSC

Figure 2. (a) LF of old GCs with Gaussian for GCs in (Milky Way + M31). (b) MF of the young clusters. pleteness limit and with (V − I) colours available, the MF is compatible with a Gaussian MF with parameters hLog(MYSC/M⊙)i = 5.6, σ ∼ 0.46 5 ←→ hMass(YSC)i∼ 4 · 10 M⊙ (Fig. 2b). If we include YSCs fainter than the completeness limit, neither the shape nor the parameters of the MF are changed (cf. F.-v.A. 1999 for details). The uncertainty in the YSC metallicity leads to age uncertainties, which, in turn, lead to uncertainties in model M/L values of the order of 10 % in these early stages. Inhomogenities in the dust distribution do not seem to be very important for the brightest YSCs for which ages from (U − V) agree with ages from (V − I) colours. We do not know, however, how important they are for those YSCs that are not detected in U. Hence, on the basis of (V − I) alone, we cannot quantify in how far an inhomogeneous internal reddening might aﬀect our results. For this reason and since our analysis is based on WFPC1 data, we caution that our conclusions concerning the evolution of the YSC LF and their MF can only be preliminary. As soon as the reduced WFPC2 data presented by Star Clusters: Metallicities, Ages, and Masses 7

Miller will become available to us, we shall repeat our analysis and supplement it with simulations to estimate the eﬀects of diﬀerential observational uncertainties.

4.3. Implications If, however, our preliminary result became conﬁrmed, this would imply that the log-normal MF of old GCs is produced by the cluster formation process rather than by secular dynamical evolution of the cluster system. In this context. it seems very interesting to obtain observational information about the molecular cloud mass spectrum in massive interacting galaxies. Jog & Solomon (1992) conjecture that the strongly enhanced ambient pressure in massive gas rich mergers might aﬀect the molecular cloud structure. In Ultraluminous Infrared Galaxies – which all are mergers with strong starbursts – the fraction of gas at very high densities of n ∼ 104 and 105 cm−3, as traced by HCN and CS lines, with respect to gas at n ∼ 500 cm−3, as traced by CO, is indeed observed to be higher by 1 – 2 orders of magnitude (e.g. Solomon et al. 1992).

4.4. Discussion Even with deeper WFPC2 data, however, the MF of the YSCs in the Anten- nae does not yet seem to be unambiguously settled. Zhang & Fall (1999) use reddening-free colour indices Q1,Q2, solar metallicity Bruzual & Charlot models, consider incompleteness and stellar contamination, and find power law MFs with slopes α ∼−2 for YSCs in the two age intervals 2.5 – 6.3 Myr and 25 – 160 Myr where Q1 and Q2 give unambiguous results. In an independent analysis of the same data, Miller (this conf.) finds a significant flattening in the observed LF at MV ∼−10.7 mag. Intriguingly, this value is exactly the turnover luminosity implied by our MF at the mean YSC age in the absence of an age spread. Miller’s method of reconstruction, however, leads to a power law YSC MF. Clearly, more work both from the modelling and observational sides is needed before we really understand the YSC MF.

4.5. Other systems For the YSCs in NGC 7252 and NGC 3921, Miller & Fall (1997) and Miller (this conf.) prefer power law MFs. Distances to those systems, however, are larger by factors 3 and 4, respectively, pushing the completeness limit to higher luminosities, even in WFPC2 data. Moreover, the YSCs in these dynamically old merger remnants have higher mean ages (650 – 750 Myr for NGC 7252, 250 – 750 Myr for NGC 3921, as compared to 200 Myr for the Antennae), and hence, are intrinsically fainter already by 0.8 – 1 mag, on average. If we assume that the MFs for the YSCs in NGC 7252 and NGC 3921 were identical to the one we derived for the YSCs in the Antennae, we obtain turn-over luminosities hMVi = −10 . . . − 9.5 mag for NGC 7252 and hMVi = −9.5 . . . − 9 mag for NGC 3921 at their respective mean YSC ages. In both cases, these turn-over luminosities are close to the completeness limits. The diﬃculty to disentangle the old GC and the YSC populations also increases with increasing YSC age. We therefore doubt that it is possible to track the MF beyond a possible turnover in NGC 7252 and 3921. In any case, the YSCs in NGC 7252 and 3921 have survived ≫ 10 crossing times and from this fact alone are young GCs rather 8 U. Fritze – v. Alvensleben that open ones, as also indicated by their combination of small eﬀective radii and high luminosities (cf. Miller et al. 1997, Schweizer et al. 1996).

4.6. Dynamical Evolution Stellar mass loss is included in our models for the evolution of M/L. Over 12 Gyr, the masses of YSCs will decrease by ∼ 10 and 15% for a Salpeter and Scalo-IMF, respectively. About half of the entire stellar mass loss occurs during the 1st Gyr. External dynamical effects from an interaction of the clusters with the potential of the interacting galaxy system seem extremely difficult to model. Comparison of YSC LFs and MFs in an age sequence of interacting galaxies, mergers, merger remnants, and dynamically young ellipticals will allow to “see these processes at work”. In the Milky Way potential, Vesperini (1998) has shown that an assumed initially log-normal GC MF is conserved in shape and parameters during self- similar evolution over a Hubble time despite the destruction of ∼ 50% of the cluster population. If, on the other hand, he starts with a power law MF, severe fine-tuning is required for his model parameters to secularly transform it into the observed log-normal MF of old GC systems. It is clearly important to analyse more YSC systems in order to see if (and in how far) their MFs are universal or might depend on environment. Old GC systems have their turn-over around hMVi∼−7.2 mag. With 10m telescopes they are accessible to more than Virgo cluster distances. MOS – e.g. with FORS on the VLT – in combination with HST imaging will allow to determine cluster abundances and hence to more precisely age-date them, and may even provide kinematic information for independent mass estimates. An open question seems to me if the (globular) cluster formation process in the high metallicity environment of interacting spirals today is the same or not as it was in the Early Universe when the radiation field was stronger and the metallicity lower. Acknowledgments. I wish to thank the organisers and in particular Ariane Lan¸con for a stimulating workshop in a very pleasant and warm atmosphere in cold and rainy Strasbourg.

References

Ashman, K. M., Conti, A., Zepf, S. E., 1995, AJ 110, 1164 Barnes, J. E., 1988, ApJ 331, 699 Bressan, A., Chiosi, C., Fagotto, F., 1994, ApJS 94, 63 Brodie, J. P., Schrder,¨ L. L., Huchra, J. P., et al. 1998, AJ 116, 691 Bruzual, G. A., Charlot, S., 1993, ApJ 405, 538 Cˆot´e, P., Welch, D. L., Fischer, P., Gebhardt, K., 1995, ApJ 454, 788 Elmegreen, B. G., Efremov, Y. N., 1997, ApJ 480, 235 Fritze – v. Alvensleben, U., 1998, A&A 336, 83 Fritze – v. Alvensleben, U., 1999, A&A 342, L25 Fritze – v. Alvensleben, U., Burkert, A., 1995, A&A 300, 58 Star Clusters: Metallicities, Ages, and Masses 9

Fritze – v. Alvensleben, U., Gerhard, O. E., 1994, A&A 285, 751 + 775 Harris, W. E., 1991, ARA&A 29, 543 Harris, W.E., Pudritz, R. E., 1994, ApJ 429, 177 Jog, C. J., Solomon, P. M., 1992, ApJ 387, 152 Kennicutt, R. C., 1989, ApJ 344, 685 Kurth, O., 1996, Diploma Thesis, Univ. G¨ottingen Kurth, O., Fritze - v. Alvensleben, U., Fricke, K. J., 1999, A&AS 138, 19 Lada, E., Bally, J., Stark, A. A., 1991, ApJ 368,432 Leitherer, C., et al. 1999, ApJS 123, 3 Leonard, P. J. T., Richer, H. B., Fahlman, G. G., 1992, AJ 104, 2104 Meurer, G. R., 1995, Nat 375, 742 Miller, B. W., Fall, S. M., 1997, AAS 119, #115.04 Moore, B., 1996, ApJ 461, L13 Scalo, J. M., 1986, Fundam. Cosm Phys. 11, 1 Schweizer, F., Seitzer, P., 1993, ApJ 417, L29 Schweizer, F., Seitzer, P., 1998, AJ 116, 2206 Schweizer, F., Miller, B. W., Whitmore, B. C., Fall, S. M., 1996, AJ 112, 1839 Solomon, P. M., Rivolo, A. R., Barrett, J., Yahil, A., 1987, ApJ 319, 730 Solomon, P. M., Downes, D., Radford, S. J. E., 1992, ApJ 387, L55 Vesperini, E., 1998, MN 299, 1019 Whitmore, B.C., Schweizer, F., 1995, AJ 109, 960 (WS95) Worthey, G., 1994, ApJS 95, 107 Zepf, S. E., Ashman, K. M., 1993, MN 264, 611 Zhang, Q., Fall, S. M., 1999, ApJ 527, L81